저자: 에릭—출연자소개명시없음,채널운영자 · 한국인
요약 보기
결론: 캐시 우드의 메모리 반도체(삼성전자·SK하이닉스) 비관론에 공감하나, 메모리 가격 급등에 따른 수요 파괴와 기술 대체 가능성을 근거로 메모리 하락 가능성을 제기하고, 헬스케어·바이오와 AI 플랫폼 관련주를 상대적으로 유망하게 봄. 메모리 반도체: 메모리 가격이 1년여 만에 4~6배 상승했다는 점을 들어, 가격 상승이 수요 파괴(demand destruction)를 일으켜 결국 가격 하락으로 이어질 수 있다고 봄. 하이퍼스케일러 CAPEX가 반도체 기업 현금흐름으로 직결되는 구조가 '시소가 한쪽으로 기운' 상태이며, 이때 기술 혁신에 의한 파괴가 쉽게 일어난다고 주장. 기술 대체 근거: Cerebras, Groq 등 올해 상장 기업들이 HBM이 필요 없는 아키텍처를 쓴다고 언급하고, 여러 AI 개발 영역에서 HBM 없이도 작동하도록 설계하는 움직임이 있다고 함. 반도체는 상품화(commodity) 자산이라 세대 간 성능 차이 외에는 차별성이 적어 취약하다고 봄. 사이클 신호: 한국 기업들이 시설 투자를 공격적으로 확대하고 미국 자본시장 접근(ADR 등)을 시도하는 것을 과잉투자와 사이클 말기 신호로 해석함. 대안 투자: 메모리 대신 고가 문제를 기술적으로 해결하는 기업(예: Cerebras)을 선호. 헬스케어/바이오는 라이트 법칙(유전체 분석 비용 2003년 약 27억 달러 → 현재 100달러 미만)과 AI 적용 확대로 성장 잠재력이 크지만, 지수 비중 축소와 규제 리스크로 저평가 상태이며 ARKG 공매도 비중이 60%를 넘는다고 언급. AI·빅테크: 플랫폼형 기업(OpenAI, Anthropic, Palantir 등)이 SaaS를 대체할 것으로 전망. 빅테크는 AI 투자 효과가 클라우드 매출 성장률에 나타나고 있어 50/50으로 봄. 로보택시·휴머노이드를 유망 분야로 제시하며 ARK 로봇 ETF 상위 종목(테슬라, 테라다인, AMD, 팔란티어, 구글, Amazon 등)을 수혜주로 언급. 캐시 우드 평가: 기술·거시 분석 능력은 높이 평가하나, 밸류에이션(EBITDA 멀티플 5년 후 기준) 적용 방식이 느슨하다고 비판. 기술과 가격 분석 간 격차를 지적하며 두 인물을 거장으로 보지 않는다고 밝힘.
다듬은 글
yes. Today's topic: Cathie Wood mentioned the Samsung Knicks. So, they said that Elon Musk would kill Samsung and the Knicks . Today's protagonist is Cash Wood, a celebrity who needs no further introduction. Well, he is not a master, but he is a famous person. So, at the end, I would like to explain why Tom, including Cathiewood, is not a master, and if so, why I consistently cover studies on masters even though he is not .
yes. Then why did I say the Knicks would fail? This is precisely because memory prices are rising. This is a chart showing memory prices. Did n't listing prices rise by almost four or six times in just over a year? But if memory prices go up, it's obviously good for memory semiconductor companies, so why is this being called bad news rather than good news?
That is precisely because of the destruction of demand . Demand destruction is a concept that frequently appears in the context of raw materials. It refers to the phenomenon where, if the price of a specific product becomes too high, demand decreases, eventually causing the price of that asset to fall .
For example, when oil prices are excessively high, people tend to avoid driving cars and instead use public transportation or purchase electric vehicles. Furthermore, if overall prices rise due to high oil costs, there is a high probability of an economic recession; consequently, a downturn in demand leads to a drop in oil prices .
In particular, it discusses the effect where demand is destroyed by price in this way, and that destroyed demand then drives prices down again . However, the point is that we are seeing that now in semiconductors, especially in memory semiconductors . This graph is what it is. This is data I have shown you a few times, but the left side shows the projected CAPEX scale for 2026 of what we call hyperscalers .
The right side shows the projected free cash flow of these semiconductor companies . Now, if you look at the red box, you can see that the money spent by hyperscalers goes straight into the pockets of semiconductor companies. But why is this a sign of the demand wave? That means the seesaw is completely tipped right now.
Since one side is unilaterally spending money while the other is unilaterally making money, the seesaw is completely tipped right now. When the seesaw tilts like this, demand destruction occurs very easily in the technology industry . Especially now, we call the technology industry 'dission.' The point is that demand disruption caused by technological innovation is occurring frequently.
For example, there is a case where Musk redesigned his batteries to reduce cobalt usage because Congolese cobalt became a problem . Likewise, regarding memory semiconductors, if you look at the architectures of companies that went public this year, such as Cerabras or Groy, it is said that they all use architectures that do not require HBM .
And if you look into the AI development fields of other companies, it is said that there is a movement to eliminate the need for HBM in that sector as well—that is, they are currently designing systems so that development works well even without HBM . So, it means that disruption caused by this technological prowess is visible.
The point is that there is such a possibility in the memory sector, and especially since semiconductors are commercialized products. So, the fact that it has been commercialized means there isn't that much difference in the product, just like bread in a bakery. However, in the case of semiconductors, while there is a significant performance difference when moving between generations, there is no hugely meaningful difference between semiconductors of the same generation .
That is why it means that commodified assets, such as semiconductors, are even more vulnerable to such disruptions . And now, there is talk that signs of the semiconductor industry's tail end have been detected. Didn't I mention earlier that this semiconductor has been commercialized ? Typically, most of these commodified assets are cyclical in nature .
We call it so-called cyclical . Capital is flowing into the industry known as ciclickers . They say this was a traditional signal of the tail end. However, signs of the tail end are now appearing. Korean companies, by their nature, tend to invest almost aggressively once they gain confidence in a specific field.
It is said that the fact that 3JNIX is issuing facility certificates very aggressively and making moves to access the U.S. capital market are signals that funds are flowing in . So, he said that rather than buying memory semiconductor companies like this, he would invest in companies that are trying to technically solve the high-price problem, such as the Cerabras I mentioned earlier .
yes. In that case, memory semiconductors aren't very good. Then, besides semiconductors, what stocks does Cathie Wood think will be beneficiaries? It was healthcare. From Cathie Wood's perspective, there are characteristics of the technology industry that are truly received multiple times. It is precisely the fact that it has deep roots.
The fact that it follows Wright's Law. It has the advantage of being versatile. Now, regarding the point of having deep roots, it is not that something suddenly popped out without any prior technique, but rather that prior techniques gradually increased and then springed forth; that is what we mean by having deep roots .
Now, Wright's Law is what we call the learning curve . It refers to an effect where costs actually decrease as production volume increases . And paradigmaticism refers to a technology being useful not only in a specific field but also applicable to various other fields . Why is this a characteristic of the truly successful technology industry?
It is said that if we compare the nature of the dot-com bubble with that of the current AI rally, we can see how these factors, particularly the "deep horn," create such a difference . In the case of the dot-com bubble, the internet did emerge, but the other technologies and infrastructure needed to bring it to fruition—that is, to foster the industry—were lacking.
Notably, there was no cloud back then. So, if the cloud industry had existed on the foundation of the internet, a new industry that generates enormous profits would have been born from that internet infrastructure, allowing the industry to grow significantly; however, at that time, such a foundation was lacking .
However, people say that because the investment sentiment of these people is very positive and, in a word, the market is hot, the internet is still in its infancy—meaning opportunities are quite limited—yet there is too much money willing to invest, resulting in a mismatch and the formation of a bubble .
So, it is said that the characteristics of the multi-bubble at that time were analyzed based on a basic 10-year period . When analyzing a certain internet company, you think, " Ah, it will grow like this in 10 years ." They said that in 10 years, the number of visitors would increase and that it would grow, but paradoxically, this was not done to look at long-term growth potential and plan for long-term investment; rather, it was done to rationalize the current situation where opportunities are scarce and other foundational technologies or infrastructure are lacking—in other words, to justify the current price, they should have looked beyond 10 years .
However, on the other hand, AI is different. AI is based on infrastructure and technology accumulated over more than 20 years . There have been many dark ages for AI . In particular, deep learning created a breakthrough in the dark age of machine learning in the early 2010s . And because we had the cloud, we were able to handle a lot of data, and because semiconductor performance improved significantly, haven't we managed to secure this level of computing power ?
So, first of all, AI has deep roots, and in addition, cases like the multi-bubble rose in a straight line. The stock price just kept going up, but now the volatility is insane. So, they say that the insane volatility is actually considered a positive factor . Because one of the reasons for the current volatility is that people are focusing solely on a short-term perspective—thinking that it is true one can make a decent profit compared to spending money on AI— and there is so much concern in the market that volatility is expanding like this.
The point is that this has the effect of preventing such volatility from overheating to some extent in advance. So, there are currently many concerns about overvaluation, but from Cathie Wood's perspective, look at entropy. Revenue was projected to be around $9 billion by the end of 2025, but in June of this year, holiday-based revenue was about $45 billion.
So, hasn't sales based on the holiday period increased by more than five times in just six months ? While it is rising explosively to this extent, the market is showing such variability due to concerns, so people are saying that AI is different from Tacomber . So, in this context, when looking at bio, the point is to look at healthcare.
The core of healthcare is multifactorial analysis, which is the emergence of technology that allows us to analyze genomes, proteomics, and similar things at a very microscopic level. However, one of the key aspects of current analysis technology—specifically biotechnology—is that it has benefited from AI .
As we saw earlier, one of the characteristics of technologies that truly benefit is versatility. In the field of biology, AI is now applied to generality, and isn't there also Wright's Law ? Wright's Law describes the effect that costs actually decrease as production increases, and in 2003, the cost to analyze one person's genome was approximately $2.7 billion.
But now it is less than $100. So, the price has dropped unbelievably over the past 20 years or so, and it is said that even now, the cost drops by 40% every time the cumulative number of genome sequences doubles . So, since costs have already dropped significantly and are expected to continue falling rapidly, it was stated that with this level of cost reduction, new demand and business models in the bio sector will become a reality .
So, the point is that things that were impossible back in 2003 will now be built into industries because the costs have become cheaper. yes. It is said that the biotech sector is currently significantly undervalued, despite possessing such potential. Now, there are many ETFs at Ark Invest, you know .
Among them, there is an ETF themed around biotech called Arc G, and its short selling ratio is currently just over 60%. There is currently a heavy concentration of negative bets, likely because the sector is so closely tied to regulations that there are regulatory risks, and also because related companies are unable to generate proper cash flow.
Ah, I hear that they view it very negatively because of issues like this. In addition to that, the industry currently uses indices as a standard . So, as passive investing has become the mainstream trend, corporate analysis focuses on companies with high weighting in the index, while companies with low weighting or those not even included in the index are considered of no interest to the market .
However, since the biotech sector currently has a low weighting, interest has dropped significantly. Also, while various M&As are actively taking place in the biotech industry, the companies involved are outside the index. This means that despite the industry's growth, the stock market is not paying much attention to it.
Also, they are spending a huge amount of money in the biotech sector, and actually, right now, it is better to spend money to be a good company. Because, whether it is AI or biotechnology, at a time when technologies are facing an inflection point, more money must be spent to ride the wave of technological innovation in order to survive.
People are saying that because the current market prioritizes spending less, the biotech sector is shining through based on these market standards and is therefore undervalued . yes. So, bye like this. I viewed Gal Square favorably, but then how are the remaining companies being handled? Okay, first of all, I mentioned SARS.
I have a different perspective on this. I said that SAS would be replaced by PAS due to data issues. So, since the client companies of SaaS firms will be reluctant to provide their own data, it stands for Service in the PAS platform. In short, I am talking about companies like OpenAI, Entropy, and Palantir.
So, it is a company that provides a platform. I predicted that it would be replaced because, unlike software, demand would shift to such companies since clients do not have to transfer data ownership. Of course, there are exceptions, but I will discuss that later if I have time. In the case of Big Tech, it is 50/50.
So, Cathie Wood has been speaking very negatively about Big Tech for a long time . But now it has changed for the better. Since you said it was 50/50 . Because Big Tech companies are investing heavily in AI to survive, and this is starting to show up in things like cloud revenue growth rates, I said that it is positive because it shows that companies are spending money to survive in the technology sector and that it is having some effect .
But of course, it’s fifty-fifty because there is a possibility that these companies could be replaced by new players like OpenAI or Entropic . So, he said that although he holds it, the weighting has changed in the top 10 . Besides healthcare, what other industries do you view favorably? I also viewed robotaxis and humanoids very favorably .
As we saw earlier, I mentioned that versatility is important among the elements of technological capability. However, the reason why paradigm shifting is important is that this technology must become a platform for it to become an industry and grow. So, what Casid mentioned—AI, robotics, blockchain, energy storage, and things like organic sequencing—can become a single platform.
Various industries can grow based on that technology. For example, 3D printing and nanotechnology are very important and innovative technologies, but they ended up becoming dependent on other industries . For example, in the case of nanotechnology, it has simply been integrated into the manufacturing process .
There is no such thing as a nano company. So, in this way, it is important to distinguish between technologies that can serve as platforms and those that cannot. This means that Robotsi and Humanoid are companies that benefit from technologies capable of becoming platforms. In the case of Robotex, robotics naturally falls under this category.
Let 's move on to energy storage. AI is coming in. So, because the three platform technologies have been combined, it is expected to create a tremendous overlap effect, becoming a business worth over $100 trillion within roughly 5 to 10 years . Humanoid said that a robot with such tremendous potential is more than that .
yes . So, what companies does Cashward view favorably regarding robots and humanoids ? You can find some hints in this Ark ETF . Now, there is an ETF called ARK QQ, which is themed around automation technology and robotics . If you look at the top 10 stocks, they are Tesla, SpaceX, Kratos, Teradyne, AMD, Palantir, Google, Loke Labs, MBD, and Amazon.
These are the top 10 stocks. KCO views these companies as beneficiaries of the robot taxis and humanoids he mentioned. yes. Finally, since Cathie Wood is a tech stock specialist, he shared his tips. Regarding Tech God, he predicted that there is a high possibility of a winner-takes-all scenario, especially with this AI.
And there is a characteristic of the technology god: it draws an S-curve. So, the characteristic of technological innovation is that it starts off quiet without any particular issues, but then at some point, it grows explosively—almost vertically. Considering this, if there is a sudden, explosive growth where, instead of all companies sharing the fruits of that growth, a few companies monopolize them, then the companies that spend money to avoid missing out on the opportunity are considered good companies at this point .
And as we saw earlier, since there is a point where explosive growth occurs and it becomes a winner-takes-all phase, I said that the tip for investing in tech stocks these days is to look at companies based on five years ahead rather than the present or one year from now . Yes, that was the story of Cathiewood.
To add a little more to our thoughts, it is as follows. I'm not a master, neither am I, Tom Lee nor Cathie Wood. I am not a master, but I keep creating content for people, and there is a reason for that. First of all, the reason these people are not masters is, of course, that their records aren't of a master level, but also because while their strengths lie in technical analysis and macroeconomics, they show significant weaknesses in the investment aspect, particularly regarding pricing.
In the case of Cathie Wood, this is a point that came up in the recent interview. He argued that while people tend to think too positively about technology and assume that he is someone who blindly buys things if they are good, he is someone who applies very strict valuations . So, the basis for that is what we call a multiple, you know.
PER, PBR, and PSR are called multiples, and among those multiples, there is one called EB. It is called an EBITDA multiple, which refers to looking at the multiple of a company's total value relative to its EBITDA . So, he says that when looking at the Evita multiple, he applies it quite tightly, around 18 to 19 times .
However, the standard for this is five years from now . So, I estimate the EBT multiple based on five years from now, and I am saying that the estimated EBT multiple must be between 18 and 19 times or lower . However, in that process, for example, we can assume that the multiple may decrease because the stock price is expected to rise explosively and Ebidan will increase more than 100 times over five years.
To put it in extreme terms . So, ultimately, looking five years ahead means that we are not approaching it conservatively. However, since he says he views this approach conservatively, it is n't very reliable in terms of pricing. Uh, for example, why did the semiconductor market decline ? Everyone thinks Ford is undervalued.
They said it was undervalued considering its future growth potential, but looking at the current figures based on TTN, it was expensive, wasn't it? So is n't that why it was adjusted? Because the very concept of future predictions was overheated. So, similarly, the cached analysis method also carries such risks, but the fact that you speak in this manner without recognizing that indicates a weakness in terms of price analysis.
You know how I always emphasize the Three-Body ratio in investing? Just like the classic Three-Body problem, investing is difficult because while analyzing the potential of a technology or the growth prospects of an industry is relatively easy, it becomes truly challenging once the element of price is involved.
It is really difficult to determine whether this is a shelf type or not, and whether it is cheap or expensive. However, since Caswood reveals a weakness here, the same goes for Tumly. So, the reason these two figures don't reach the ranks of masters is that, while there are weaknesses in bringing in their stories, the ideas are decent and quite reasonable when it comes to technical or macro analysis.
Wouldn't you say the talk about the Knicks in this third series is also reasonable ? So, because of this aspect, I often bring up the story of Topley and Cathie Wood . So, regarding the discussion on Cathie's Wood today, the criticism that it destroys demand is reasonable, especially in the case of the Samsungnix sector.
Well, that's reasonable. In that case, regarding the likelihood of the disruption of classes suggested by CASD occurring, for example, considering the current level of technological advancement and commercialization of the companies CASD mentioned, there are cases where disruption has occurred because the pie has grown so large that the supply has increased significantly, leading other companies to jump in.
So, I would like to suggest that it is best to utilize this method to check such scenarios as well. That concludes today's content. So, let's get into the Q&A .
Comments 0